++
++
A meaningful analysis of fluid, electrolyte, and acid–base abnormalities is dependent on the history and physical examination of each patient. Although a rigorous appraisal of laboratory parameters often yields the correct differential diagnosis, the clinical characteristics provide an understanding of the extracellular fluid volume (ECFV) and pathophysiology. Thus, the evaluation always begins with an overall assessment of the patient.
+++
INITIAL PATIENT ASSESSMENT
++
The history should be directed toward clinical questions associated with fluid and electrolyte abnormalities. Xenobiotic exposure commonly results in fluid losses through the respiratory system (hyperpnea and tachypnea), gastrointestinal (GI) tract (vomiting and diarrhea), skin (diaphoresis), and kidneys (polyuria). Patients with ECFV depletion often complain of dizziness, thirst, and weakness. Usually the patients can identify the source of fluid loss.
++
A history of exposure to nonprescription and prescription medications, alternative or complementary therapies, and other xenobiotics can suggest the most likely electrolyte or acid–base abnormality. In addition, patient characteristics (race, gender, age), premorbid medical conditions, and the ambient temperature and humidity should always be considered.
++
The vital signs are invariably affected by significant alterations in ECFV. Whereas hypotension and tachycardia often characterize life-threatening ECFV depletion, an increase of the heart rate and a narrowing of the pulse pressure are the earliest findings. Abnormalities are best recognized through an ongoing dynamic evaluation, realizing that the measurement of a single set of supine vital signs offers useful information only when markedly abnormal. Orthostatic pulse and blood pressure measurements provide a more meaningful determination of functional ECFV status (Chaps. 3 and 16).
++
The respiratory rate and pattern can give clues to the patient’s metabolic status. In the absence of lung disease, hyperventilation (manifested by tachypnea, hyperpnea, or both) is often either caused by a primary respiratory stimulus (respiratory alkalosis) or is a response to the presence of metabolic acidosis. Although hypoventilation (bradypnea or hypopnea or both) is present in patients with metabolic alkalosis, it is rarely clinically significant except in the presence of chronic lung disease or in combination with respiratory depressant xenobiotics. More commonly, hypoventilation is caused by a primary depression of consciousness and respiration with resulting respiratory acidosis. Unless the clinical scenario (eg, nature of the poisoning, presence of renal or pulmonary disease, and findings on physical examination or laboratory testing) is classic, arterial or venous blood gas analysis is recommended to determine the acid–base disorder associated with a change in ventilation.
++
The skin should be evaluated for turgor, moisture, and the presence or absence of edema. The moisture of the mucous membranes can also provide valuable information. These are nonspecific parameters and often fail to correlate directly with the status of hydration. This dissociation is especially true with xenobiotic exposure because many xenobiotics alter skin and mucous membrane moisture without necessarily altering ECFV status. For example, antimuscarinics commonly result in dry mucous membranes and skin without producing ECFV depletion. Conversely, patients exposed to sympathomimetics such as cocaine or cholinergics such as organic phosphorus compounds have moist skin and mucous membranes even in the presence of significant fluid losses. This dissociation of ECFV and cutaneous characteristics reinforces the need to assess patients meticulously.
++
The physical findings associated with electrolyte abnormalities are generally nonspecific. Hyponatremia, hypernatremia, hypercalcemia, and hypermagnesemia all produce a depressed level of consciousness. Neuromuscular excitability such as tremor and hyperreflexia occurs with hypocalcemia, hypomagnesemia, hyponatremia, and hyperkalemia. Weakness results from both hyperkalemia and hypokalemia. Also, multiple, concurrent electrolyte disorders can produce confusing clinical presentations, or patients can appear normal. Diagnostic findings, such as Chvostek and Trousseau signs (primarily found in hypocalcemia), are useful in assessing patients with potential xenobiotic exposures.
+++
Rapid Diagnostic Tools
++
The electrocardiogram (ECG) is a useful tool for screening several common electrolyte abnormalities (Chap. 15). It is easy to perform, rapid, inexpensive, and routinely available. Unfortunately, because poor sensitivity (0.43) and moderate specificity (0.86) were demonstrated when ECGs were used to diagnose hyperkalemia, in actuality the test is of limited diagnostic value.151 However, the ECG is valuable for the evaluation of changes in serum potassium and calcium concentrations ([K+] and [Ca2+]) in an individual patient.
++
Assessment of urine specific gravity by dipstick analysis provides valuable information about ECFV status. A high urine specific gravity (>1.015) signifies concentrated urine and is often associated with ECFV depletion. However, urine specific gravity is often similarly elevated in conditions of ECFV excess, such as congestive heart failure or third spacing. Similarly, patients with excess antidiuretic hormone (ADH) secretion (eg, due to methylenedioxymethamphetamine {MDMA} exposure) excrete concentrated urine (specific gravity >1.015) even in the presence of a normal or expanded ECFV. Furthermore, when renal impairment or diuretic use is the source of the volume loss, the specific gravity is usually approximately 1.010 (known as isosthenuria). Finally, patients with lithium-induced diabetes insipidus (DI) excrete dilute urine (specific gravity <1.010) despite ongoing water losses and contraction of the ECFV.
++
The urine dipstick is highly reliable and rapidly available for determining the presence of ketones, which are often associated with common causes of metabolic acidosis (eg, diabetic ketoacidosis, salicylate poisoning, alcoholic ketoacidosis). The urine ferric chloride test rapidly detects exposure to salicylates with a high sensitivity and specificity although it is rarely used today in locations where salicylate concentrations can be obtained rapidly (Chap. 37).
++
A simultaneous determination of the venous serum electrolytes, blood urea nitrogen (BUN), glucose, and arterial or venous blood gases are adequate to determine the nature of the most common acid–base, fluid, and electrolyte abnormalities. More complex clinical problems require determinations of urine and serum osmolalities, urine electrolytes, serum ketones, serum lactate, and other tests. A systematic approach to common problems is discussed in the following sections.
+++
ACID–BASE ABNORMALITIES
++
The lack of a clear understanding and precise use of the terminology of acid–base disorders often leads to confusion and error. The following definitions provide the appropriate frame of reference for the remainder of the chapter and this textbook.
++
Whereas the terms acidosis and alkalosis refer to processes that tend to change pH in a given direction, acidemia and alkalemia only refer to the actual pH. By definition, a patient is said to have:
++
A metabolic acidosis if the arterial pH is less than 7.40 and serum bicarbonate concentration ([HCO3−]) is less than 24 mEq/L. Because acidemia stimulates ventilation (respiratory compensation), metabolic acidosis is usually accompanied by a PCO2 less than 40 mm Hg.
A metabolic alkalosis if the arterial pH is greater than 7.40 and serum [HCO3−] is greater than 24 mEq/L. Because alkalemia inhibits ventilation (respiratory compensation), metabolic alkalosis is usually accompanied by a PCO2 greater than 40 mm Hg.
A respiratory acidosis if the arterial pH is less than 7.40 and partial pressure of carbon dioxide (PCO2) is greater than 40 mm Hg. Because an elevated PCO2 stimulates renal acid excretion and the generation of HCO3− (renal compensation), respiratory acidosis is usually accompanied by a serum [HCO3−] greater than 24 mEq/L.
A respiratory alkalosis if the arterial pH is greater than 7.40 and PCO2 is less than 40 mm Hg. Because a decreased PCO2 lowers renal acid excretion and increases the excretion of HCO3− (renal compensation), respiratory alkalosis is usually accompanied by a serum [HCO3−] less than 24 mEq/L.
++
It is important to note that under many circumstances, a venous pH permits an approximation of arterial pH. However, when rigorously compared, the arterial blood gas analysis outperforms the venous blood gas analysis especially when findings are subtle.131 (See Chap. 28 for a further discussion of the relationship between arterial and venous pH.) Any combination of acidoses and alkaloses can be present in any one patient at any given time. The terms acidemia and alkalemia refer only to the resultant arterial pH of blood (acidemia referring to a pH <7.40 and alkalemia referring to a pH >7.40). These terms do not describe the processes that led to the alteration in pH. Thus, a patient with acidemia must have a primary metabolic or respiratory acidosis but potentially also has an alkalosis present at the same time. Clues to the presence of more than one acid–base abnormality include the clinical presentation, an apparent excess or insufficient “compensation” for the primary acid–base abnormality, a delta gap ratio {(Δ) anion gap/Δ[HCO3−]} that significantly deviates from one, or an electrolyte abnormality that is uncharacteristic of the primary acid–base disorder.
+++
Determining the Primary Acid–Base Abnormality
++
It is helpful to begin by determining whether the patient has an acidosis or an alkalosis. This is followed by an assessment of the pH, PCO2, and [HCO3−]. With these 3 parameters defined, the patient’s primary acid–base disorder can be classified using the aforementioned definitions. Next it is important to determine whether the compensation of the primary acid–base disorder is appropriate. It is generally assumed that overcompensation cannot occur.106 That is, if the primary process is metabolic acidosis, respiratory compensation tends to raise the pH toward normal but never to greater than 7.40. If the primary process is respiratory alkalosis, compensatory renal excretion of HCO3− tends to lower the pH toward normal but not to less than 7.40. The same is true for primary metabolic alkalosis and primary respiratory acidosis. As a rule, compensation for a primary acid–base disorder that appears inadequate or excessive is indicative of the presence of a second primary acid–base disorder.
++
Based on patient data, the Winters equation (Equation 12–1)7 predicts the degree of the respiratory compensation (the decrease in PCO2) in metabolic acidosis as follows:
++
++
Thus, in a patient with an arterial [HCO3−] of 12 mEq/L, the predicted PCO2 is calculated as
++
++
++
++
If the actual PCO2 is substantially lower than is predicted by the Winters equation, it can be concluded that both a primary metabolic acidosis and a primary respiratory alkalosis are present. If the PCO2 is substantially higher than the predicted value, then both a primary metabolic acidosis and a primary respiratory acidosis are present.
++
An alternative to the Winters equation is the observation by Narins and Emmett that in compensated metabolic acidosis, the arterial PCO2 is usually the same as the last 2 digits of the arterial pH.106 For example, a pH of 7.26 predicts a PCO2 of 26 mm Hg.
++
Guidelines are also available to predict the compensation for metabolic alkalosis,64 respiratory acidosis, and respiratory alkalosis.83 Patients with a metabolic alkalosis compensate by hypoventilating, resulting in an increase of their PCO2 above 40 mm Hg. However, the concomitant development of hypoxemia limits this compensation so that respiratory compensation in the presence of a metabolic alkalosis usually results in a PCO2 of 55 mm Hg or less. It is difficult to be more accurate about the expected respiratory compensation for a metabolic alkalosis, although the compensation, as in the case of metabolic acidosis, is nearly complete within hours of onset.
++
By contrast, the degree of compensation in primary respiratory disorders depends on the length of time the disorder has been present. In a matter of minutes, primary respiratory acidosis results in an increase in the serum [HCO3−] of 0.1 times the increase in the PCO2. This increase is a result of the production and dissociation of H2CO3. Over a period of days, respiratory acidosis causes the compensatory renal excretion of acid. This compensation increases the serum [HCO3−] by 0.3 times the increase in PCO2. Primary respiratory alkalosis acutely decreases the serum [HCO3−] by 0.2 times the decrease in PCO2. If a respiratory alkalosis persists for several days, renal compensation, by the urinary excretion of HCO3−, decreases the serum [HCO3−] by 0.4 times the decrease in PCO2.
+++
Calculating the Anion Gap
++
The concept of the anion gap is said to have arisen from the “Gamblegram” originally described in 193951; however, its use was not popularized until the determination of serum electrolytes became routinely available. The law of electroneutrality states that the net positive and negative charges of all fluids must be equal. Thus, all of the negative charges present in the serum must equal all of the positive charges, and the sum of the positive charges minus the sum of the negative charges must equal zero. The problem that immediately arose (and produced an “anion gap”) was that all charged species in the serum are not routinely measured.
++
Normally present but not routinely measured cations include calcium and magnesium; normally present but not routinely measured anions include phosphate, sulfate, albumin, and organic anions (eg, lactate and pyruvate).41 Whereas Na+ and K+ normally account for 95% of extracellular cations, Cl− and HCO3− account for 85% of extracellular anions. Thus, because more cations than anions are among the electrolytes usually measured, subtracting the anions from the cations normally yields a positive number, known as the anion gap. The anion gap is therefore derived as shown in Equation 12–2:
++
++
++
++
Because the serum [K+] varies over a limited range of perhaps 1 to 2 mEq/L above and below normal and therefore rarely significantly alters the anion gap, it is often deleted from the equation for simplicity. Most prefer this approach, yielding Equation 12–3:
++
++
Using Equation 12–3, the normal anion gap was initially determined to be 12 ± 4 mEq/L.41 However, because the normal serum [Cl−] is higher on current laboratory instrumentation, the current range for a normal anion gap is 7 ± 4 mEq/L.149
++
A variety of pathologic conditions result in a rise or fall of the anion gap. High anion gaps result from increased presence of unmeasured anions or decreased presence of unmeasured cations (Table 12–1).41,89 Conversely, a low anion gap results from an increase in unmeasured cations or a decrease in unmeasured anions (Table 12–2).41,49,59,129
++
++
+++
Anion Gap Reliability
++
Several authors have discussed the usefulness of the anion gap determination.21,50,72 When 57 hospitalized patients were studied to determine the cause of elevated anion gaps in patients whose anion gap was greater than 30 mEq/L, the cause was always a metabolic acidosis with elevations of lactate or ketones.50 In patients with smaller elevations of the anion gap, the ability to define the cause of the elevation diminished; in only 14% of patients with anion gaps of 17 to 19 mEq/L could the cause be determined. Another study determined that although the anion gap is often used as a screening test for hyperlactatemia (as a sign of poor perfusion), only patients with the highest serum lactate concentrations had elevated anion gaps.72 Finally, in a sample of 571 patients, those with greater elevations in anion gaps tended to have more severe illness. This logically correlated with higher admission rates, a greater percentage of admissions to intensive care units, and a higher mortality rate.21 Thus, although the absence of an increased anion gap does not exclude significant illness, a very elevated anion gap can generally be attributed to a specific cause (typically a disorder that is associated with elevated lactate or ketones) and usually indicates a relatively severe illness.
++
After the diagnosis of metabolic acidosis is established by finding an arterial pH less than 7.40, [HCO3−] <24 mEq/L, and PCO2 <40 mm Hg, the serum anion gap should be analyzed. Indeed, the popularity of the anion gap is primarily based on its usefulness in categorizing metabolic acidosis as being of the high anion gap or normal anion gap type. This determination should be made after correcting the anion gap for the effect of hypoalbuminemia, a common and important confounding factor in chronically ill patients. The anion gap decreases approximately 3 mEq/L per 1 g/dL decrease in the serum [albumin].47 In general, the electrolyte abnormalities that frequently accompany metabolic acidosis usually have only small and insignificant effects on the anion gap.
++
It should be noted that although many clinicians rely on the mnemonics MUDPILES (M, methanol; U, uremia; D, diabetic ketoacidosis; P, paraldehyde; I, iron; L, lactic acidosis; E, ethylene glycol; and S, salicylates) or KULT (K, ketones; U, uremia; L, lactate; T, toxins), to help remember this differential diagnosis, these mnemonics include rarely used drugs (phenformin, paraldehyde) and omit important others (eg, metformin, cyanide).
++
A high anion gap metabolic acidosis results from the absorption or generation of an acid that dissociates into an anion other than Cl− that has neither been excreted nor metabolized at the time the anion gap is determined. The retention of this “unmeasured” anion (eg, glycolate in ethylene glycol poisoning) increases the anion gap. By contrast, a normal anion gap metabolic acidosis results from the absorption or generation of an acid that dissociates into H+ and Cl−. In this case, the “measured” Cl− is retained as HCO3−, is titrated, and its concentration reduced during the acidosis, and no increase in anion gap is produced. Normal anion gap acidosis, also referred to as hyperchloremic metabolic acidosis, is typically caused by intestinal or renal bicarbonate losses as in diarrhea or renal tubular acidosis, respectively. Other causes of high and normal anion gap metabolic acidoses are described elsewhere3,4 and shown in Tables 12–3 and 12–4.
++
++
+++
Narrowing the Differential Diagnosis of a High Anion Gap Metabolic Acidosis
++
The ability to diagnose the etiology of a high anion gap metabolic acidosis is an essential skill in clinical medicine. The following discussion provides a rapid and cost-effective approach to the problem. As always, the clinical history and physical examination provide essential clues to the diagnosis. For example, iron poisoning is virtually always associated with significant GI symptoms, the absence of which essentially excludes the diagnosis (Chap. 45). Furthermore, when iron overdose is suspected, an abdominal radiograph often shows the presence of iron-containing tablets. The acidosis associated with isoniazid (INH) toxicity results from seizures, the absence of which excludes INH as the cause of a metabolic acidosis (Chap. 56). Methanol poisoning is classically associated with visual complaints or abnormal finding on funduscopic examination (Chap. 106). Methyl salicylate has a characteristic wintergreen odor (Chap. 37). When these findings are absent, the laboratory analysis must be relied on, as follows:
++
Begin with the serum electrolytes, BUN, creatinine, and glucose. A rapid blood glucose reagent test should be performed to help confirm or exclude hyperglycemia. Although hyperglycemia should raise the possibility of diabetic ketoacidosis, the absence of an elevated serum glucose does not exclude the possibility of euglycemic diabetic ketoacidosis,11,27,74 alcoholic or starvation ketoacidosis, which are often associated with normal or even low serum glucose concentrations. An elevated BUN and creatinine are essential to diagnose acute or chronic kidney failure.
Proceed to the urinalysis. If there is a suspicion of a high anion gap metabolic acidosis and only the arterial or venous blood gas analysis is completed, the evaluation can easily begin here while the electrolyte determinations are pending. A urine dipstick for glucose and ketones helps with the diagnosis of diabetic ketoacidosis and other ketoacidoses. However, the absence of urinary ketones does not exclude a diagnosis of alcoholic ketoacidosis48 (Chap. 76), and ketones are often present in patients with severe salicylism (Chap. 37) and biguanide-associated metabolic acidosis (Chap. 47). If timed properly, the urine of a patient who has ingested fluorescein-containing antifreeze (ethylene glycol) fluoresces when exposed to a Wood lamp. Also, because ethylene glycol is metabolized to oxalate, calcium oxalate crystals are present in the urine of approximately half of poisoned patients. Although the presence of a fluorescent urine and calcium oxalate crystals are useful findings, their absence does not exclude ethylene glycol poisoning (Chap. 106). When clinically available, a urine ferric chloride test should be performed. Although highly sensitive and specific for the presence of salicylates, this test is not specific for the diagnosis of salicylism because small amounts of salicylate will be detected in the urine even days after its last use (Chap. 37). Thus, a serum salicylate concentration must be obtained to quantify the findings of a positive urine ferric chloride test result. A negative urine ferric chloride test result essentially excludes a diagnosis of salicylism.
A blood lactate concentration can be helpful. In theory, if the lactate (measured in mEq/L) can entirely account for the fall in serum [HCO3−], the cause of the high anion gap can be attributed to lactic acidosis. In practicality we know that this unfortunately does not work largely because the volumes of distribution of bicarbonate and lactate are not identical, with the bicarbonate volume of distribution being very dependent on pH.119 Another important example is that glycolate (a metabolite of ethylene glycol) can produce a false-positive elevation of the lactate concentration with many current laboratory techniques.100,150
++
When the above analysis of a high anion gap metabolic acidosis is nondiagnostic, the diagnosis is usually toxic alcohol ingestion, starvation, or alcoholic ketoacidosis (with minimal urine ketones), or a multifactorial process involving small amounts of lactate and other anions. One approach is to provide the patient with 1 to 2 hours of intravenous (IV) hydration, dextrose, and thiamine. If the acidosis improves, the etiology is either ketoacidosis or metabolic acidosis with hyperlacatemia. In the absence of improvement, a more detailed search for the toxic alcohols, involving measurement of either the osmol gap or actual methanol and ethylene glycol concentrations, should be initiated (discussed later).
+++
The Δ Anion Gap-to-Δ[HCO3−] Ratio
++
Many patients have mixed acid–base disorders such as metabolic acidosis and respiratory alkalosis. Depending on the relative effects of the acid–base disorders, the patient may have significant acidemia or alkalemia, minor alterations in pH, or even a normal pH. Although the clinical presentation, degree of compensation for the primary acid–base disorder, or the presence of unexpected electrolyte abnormalities suggest whether more than one primary acid–base disorder is present, comparing the Δ anion gap (ΔAG) with the Δ[HCO3−] provides additional information to help establish the correct diagnosis.
++
In a patient with a simple high anion gap metabolic acidosis, each 1 mEq/L decrease in the serum [HCO3−] should (at least initially) be associated with a 1 mEq/L rise in the anion gap.106 This occurs because the unmeasured anion is paired with the acid that is titrating the HCO3−. Any deviation from this direct relationship may be an indication of a mixed acid–base disorder.60,106,111 Thus, the ratio of the change in the anion gap (ΔAG) to the deviation of the serum [HCO3−] from normal (Equation 12–4) evolved:
++
++
A ratio close to one would suggest a pure high anion gap metabolic acidosis. When the ratio is greater than one, there is a relative increase in [HCO3−] that can result only from a concomitant metabolic alkalosis or renal compensation such as renal generation of HCO3− for a respiratory acidosis. Alternatively, when the ratio is less than one, the additional presence of either hyperchloremic (normal anion gap) metabolic acidosis or compensated respiratory alkalosis is suggested. Although the usefulness of this relationship has been supported strongly by some authors,109,111 others suggest that it is often flawed and frequently misleading.34,125
++
After reviewing the arguments, the statements of one author34 are reasonable in concluding that “the exact relationship between the ΔAG and Δ[HCO3−] in a high anion gap metabolic acidosis is not readily predictable and deviation of the ΔAG/Δ[HCO3−] ratio from unity does not necessarily imply the diagnosis of a second acid–base disorder.” Regardless, very large deviations from a value of one usually are associated with the presence of a second primary acid–base disorder.
++
The osmol gap, which is sometimes used to screen for toxic alcohol ingestion, is defined as the difference between the values for the measured serum osmolality and the calculated serum osmolarity. Osmolarity is a measure of the total number of particles in one liter of solution. Osmolality differs from osmolarity in that it represents the number of particles per kilogram of solution. Thus, osmolarity and osmolality represent molar and molal concentrations of solutes, respectively. In clinical medicine, osmolarity is usually calculated whereas osmolality is measured.
++
Calculating osmolarity requires a summing of the known particles in solution. Because molarity and milliequivalents are particle-based measurements, unlike weight or concentration, the known constituents of serum that are measured in the latter units (such as mg/dL) have to be converted to molar values. Assumptions are required based on the extent of dissociation of polar compounds (eg, NaCl), the water content of serum, and the contributions of various other solutes such as Ca2+ and Mg2+. The nature and limitations of these assumptions are beyond the scope of this chapter. Readers are referred to several reviews for more details.67,110 Many equations have been used and evaluated for calculating serum osmolarity. One investigation that used 13 different methods to evaluate sera from 715 hospitalized patients36 concluded that Equation 12–5 provided the most accurate calculation:
++
++
Obvious sources of potential error in this calculation include laboratory error in determining the measured parameters and the failure to account for a number of osmotically active particles.
++
The measurement of serum osmolality also has the potential for error stemming from the use of different laboratory technique.40 It is essential to ensure that the freezing point depression technique or osmometry is used because when the boiling point elevation method is used, xenobiotics with low boiling points (ethanol, isopropanol, methanol) will not be detected.
++
Conceptual errors may also result. In methanol poisoning, the methanol molecule has osmotic activity that is measured but not calculated, and no increase in the anion gap is present until it is metabolized to formate. Although the metabolite also has osmotic activity, its activity is accounted for by Na+ in the osmolarity calculation because it is largely dissociated, existing as Na+ formate. Thus, shortly after a methanol ingestion, the patient will have an elevated osmol gap and a normal anion gap; later, the anion gap will increase, and the osmol gap will decrease (Fig. 106–4). This effect is highlighted by several case reports.9,30
++
Using Equation 12–5 to calculate osmolarity, it is often stated that the “normal” osmol gap is 10 ± 6 mOsm/L.36 However, when more than 300 adult samples were studied with a more commonly used equation (Eq. 12–6),
++
++
normal values for the osmol gap were –2 ± 6 mOsm/L.67 Almost identical results are reported in children.96
++
The largest limitation of the osmol gap calculation is due to the documented large standard deviation around a small “normal” number.36,67 An error of 1 mEq/L (<1.0%) in the determination of the serum [Na+] will result in an error of 2 mOsm/L in the calculation of the osmol gap. Considering this variability, the molecular weights (MWs) and relatively modest serum concentrations of the xenobiotics in question (eg, ethylene glycol; MW, 62 Da; at a concentration of 50 mg/dL theoretically contributes only 8.1 mOsm/L) and the predicted fall in the osmol gap as metabolism occurs, small or even negative osmol gaps can never be used to exclude toxic alcohol ingestion.67 This overall concept is illustrated by an actual patient with an osmol gap of 7.2 mOsm (well within the normal range) who ultimately required hemodialysis for severe ethylene glycol poisoning.138 An additional error may result when including ethanol in the determination of the osmol gap. When present, ethanol is osmotically active and should be included in the calculated osmolarity. In theory, because the MW of ethanol is 46 g/mol, dividing the serum ethanol concentration (in mg/dL) by 4.6 will yield the osmolar contribution in mmol/L. However, because the physical interaction of ethanol with water is complex, it is more scientifically accurate to divide by lower numbers as the ethanol concentration increases.26,115 Therefore because the osmol gap is a screening tool, we suggest continuing to use the 4.6 divisor (or if in SI units using the unmodified molar concentration) in an attempt to reduce clinical false-negative test results.
++
Finally, although exceedingly large serum osmol gaps are suggestive of toxic alcohol ingestions, common conditions such as alcoholic ketoacidosis, metabolic acidosis with elevated lactate, kidney failure, and shock are all associated with elevated osmol gaps.73,128,134 This may be surprising because lactate, acetoacetate, and β-hydroxybutyrate should not account for any increase in the osmol gap because they are charged (and accounted for in the osmolarity calculation). Apparently, these conditions are associated with the accumulation of small uncharged, unmeasured molecules in the serum.
++
Thus, although the negative and positive predictive values of the osmol gap are too poor to recommend this test to routinely screen for xenobiotic ingestion, the presence of very high osmol gaps (>50–70 mOsm/L) usually indicates a diagnosis of toxic alcohol ingestion (Chap. 106).
+++
Differential Diagnosis of a Normal Anion Gap Metabolic Acidosis
++
Although the differential diagnosis of a normal anion gap metabolic acidosis is extensive (Table 12–4), most cases result from either urinary or GI HCO3− losses: renal tubular acidosis (RTA) or diarrhea, respectively. A number of xenobiotics also cause this disorder, including toluene,25 which also may cause a high anion gap metabolic acidosis. When the findings of the history and physical examination cannot be used to narrow the differential diagnosis, the use of a urinary anion gap is suggested.15,120
++
The urinary anion gap is calculated as shown in Equation 12–7:
++
++
The size of this gap is inversely related to the urinary ammonium (NH4+) excretion.58 As NH4+ elimination increases, the urinary anion gap decreases and can actually become negative because NH4+ serves as an unmeasured urinary cation and is predominantly accompanied by Cl−.
++
The normal anion gap metabolic acidosis associated with diarrhea results from HCO3− loss. During this process, the ability of the kidney to eliminate NH4+ is undisturbed; in fact, it increases as a normal response to the acidemia. Thus, with gastrointestinal HCO3− losses, the urinary anion gap should decrease and may become negative. By contrast, a patient with RTA has lost the ability to either reabsorb HCO3− (type 2 RTA) or increase NH4+ excretion in response to metabolic acidosis (types 1 and 4 RTA) and the urinary anion gap should become more positive. Indeed, when the urinary anion gap was calculated in patients with diarrhea or RTA, it was found that patients with diarrhea had a mean negative gap (–20 ± 5.7 mEq/L) compared with a positive gap (23 ± 4.1 mEq/L) in those with RTA.58 Therefore, when evaluating the patient with a normal anion gap metabolic acidosis, the determination of a urinary anion gap is helpful to determine the source of the disorder when the history and physical examination are unclear.
+++
Adverse Effects of Metabolic Acidosis
++
The acuity of onset and severity of metabolic acidosis determine the consequences of this disorder. Acute metabolic acidosis is usually characterized by obvious hyperventilation (caused by respiratory compensation). At arterial pH values less than 7.20, cardiac and central nervous system abnormalities are often present. These include decreases in blood pressure and cardiac output, cardiac dysrhythmias, and progressive obtundation.3,4 Chronic metabolic acidosis is often not accompanied by overt clinical symptoms. The nonspecific symptoms of anorexia and fatigue are the most typical manifestations of chronic acidosis, and compensatory hyperventilation although present is often not evident. Because the consequences of even severe metabolic acidosis are nonspecific, the presence of metabolic acidosis is most often suggested by the history and physical examination and subsequently confirmed by laboratory testing.
+++
Management Principles in Patients with Metabolic Acidosis
++
The treatment of metabolic acidosis depends on its severity and cause. In most cases of severe poisoning, with a serum [HCO3−] concentration less than 8 mEq/L and an arterial pH value less than 7.20, we recommend treating with HCO3− to increase the pH to greater than 7.20, as described in detail elsewhere.3,4 As an example, to raise the serum [HCO3−] by 4 mEq/L in a 70-kg person with an apparent HCO3− distribution space of 50% of body weight, approximately 140 mEq must be administered. Unfortunately, because the apparent volume of distribution of HCO3− increases as the pH and serum [HCO3−] fall, any given dose of exogenous HCO3− will have less of an effect on pH. When ECFV overload (caused by heart failure, kidney failure, or the sodium bicarbonate therapy itself) cannot be prevented or managed by administering loop diuretics, hemofiltration or hemodialysis will be necessary.
++
In patients with arterial pH values greater than 7.20, the cause of the acidosis should guide therapy. Metabolic acidosis primarily caused by the overproduction of acid, as in the case of ketoacidosis and toxic alcohol poisoning, requires very large quantities of HCO3− and typically does not respond well to sodium bicarbonate therapy. Treatment in these patients should be directed at the cause of acidosis (eg, insulin and IV fluids in diabetic ketoacidosis; fomepizole in methanol, ethylene glycol, and DEG poisonings (Antidotes in Depth: A33), fluids, dextrose, and thiamine in alcoholic ketoacidosis; fluid resuscitation, antibiotics, and vasopressors in sepsis-induced hyperlactatemia). Patients with metabolic acidosis primarily caused by insufficient excretion of acid (eg, acute or chronic kidney failure, RTA) should be treated with a low-protein diet (if feasible) and oral sodium bicarbonate or substances that generate HCO3− during metabolism. We recommended an oral sodium citrate solution such as Shohl solution, which yields 1 mEq base/mL. The goal of therapy is to increase the serum [HCO3−] to 20 to 22 mEq/L and the pH to 7.30.
+++
Adverse Effects of Metabolic Alkalosis
++
Life-threatening metabolic alkalosis is rare but can result in tetany (from decreased ionized [Ca2+]); weakness (from decreased serum [K+]); or altered mental status leading to coma, seizures, and cardiac dysrhythmias. In addition, metabolic alkalosis shifts the oxyhemoglobin dissociation curve to the left, impairing tissue oxygenation (Chap. 28). The expected compensation for a metabolic alkalosis is hypoventilation and increased PCO2. As discussed before, respiratory compensation is inadequate at best, invoking the teleological argument that hypoxia is more undesirable than alkalemia.106 Several authors, however, have reported that severe hypoventilation and respiratory failure can occur in response to metabolic alkalosis, suggesting an actual, although uncommon, risk.112
+++
Approach to the Patient with Metabolic Alkalosis
++
Metabolic alkalosis results from GI or urinary loss of acid, administration of exogenous base, or renal HCO3− retention (ie, impaired renal HCO3− excretion). Table 12–5 lists the causes of metabolic alkalosis. Compared with metabolic acidosis, metabolic alkalosis is less common and less frequently a consequence of xenobiotic exposure.
++
++
The etiologies of metabolic alkalosis are classically characterized from a therapeutic standpoint as chloride responsive or chloride resistant. Chloride-responsive etiologies such as diuretic use, vomiting, nasogastric suction, and Cl− diarrhea are usually associated with a urinary [Cl−] <10 mEq/L.64 Patients with these disorders respond rapidly to infusion of 0.9% NaCl solution when concomitant therapy addresses the underlying problem.3,4 Chloride resistant disorders exemplified by hyperaldosteronism and severe K+ depletion are characterized by urinary [Cl−] >10 mEq/L and tend to be resistant to 0.9% NaCl solution therapy.53,64 Patients with these disorders often require K+ repletion or drugs that reduce mineralocorticoid effects, such as spironolactone or eplerenone, before correction can occur.53 When 0.9% NaCl solution repletion is ineffective or emergent correction of the alkalosis is required, some authors have suggested infusions of lysine or arginine HCl or dilute HCl.3,4 However, this technique is rarely necessary and we do not routinely recommend it.
+++
XENOBIOTIC-INDUCED AND OTHER ALTERATIONS OF WATER BALANCE
++
Significant fluid abnormalities commonly occur following xenobiotic exposure. Gastrointestinal losses in the form of vomiting, diarrhea, GI hemorrhage, and third spacing such as from GI burns result from a variety of xenobiotic toxicities and their management with emetics and cathartics. Renal fluid losses result from the ability of many xenobiotics to increase the glomerular filtration rate (inotropes), impair Na+ reabsorption (diuretics), or increase urine volume in response to an obligate solute load (salicylates). Fluid losses also occur through the skin as a result of sweating (sympathomimetics, cholinergics, and uncouplers of oxidative phosphorylation) or the lung as a result of increased minute ventilation (salicylates and sympathomimetics) or bronchorrhea (cholinergics). To the extent that these lost fluids contain Na+, various signs, symptoms, and laboratory evidence of ECFV depletion develop.
++
The diagnosis and treatment of abnormal serum electrolyte concentrations are usually addressed after repletion of the ECFV deficit with isotonic, Na+-containing fluids (eg, blood products, 0.9% NaCl solution, lactated Ringer solution). Other fluid balance issues are discussed in Chaps. 16 and 27 and other chapters relating to individual xenobiotics. This section focuses on body water balance (abnormalities of which manifest as hypernatremia and hyponatremia) and specifically on the toxicologically relevant syndromes of diabetes insipidus (DI) and the syndrome of inappropriate secretion of antidiuretic hormone (SIADH).
++
Sodium concentration in the extracellular space is intrinsically related to and directly reflects total body water balance. This occurs because the sodium cation is largely restricted to the extracellular space, and its serum concentration is primarily, if indirectly, controlled by factors that control water balance. Thus, both the serum [Na+] and plasma osmolality vary inversely with changes in the quantity of body water.
++
Plasma osmolality is maintained through a complex interaction between dietary water intake; the hypothalamus, pituitary gland, and kidney; and the effects of hormones such as arginine vasopressin (ADH) and adrenal mineralocorticoids.17,20,108,146 Briefly, changes in osmolality are caused by changes in water intake and insensible (dermal, respiratory, and stool) and sensible (urinary, sweat) water losses. Urinary water losses are controlled by ADH. Increases in the osmolality of the extracellular fluid (ECF) stimulate anterior hypothalamic osmoreceptors, thereby stimulating thirst and ADH synthesis and release by the posterior pituitary gland. Arginine vasopressin release reaches its maximum concentration at a plasma osmolality of about 295 mOsm/kg. Arginine vasopressin is transported to the kidney via the bloodstream, where it stimulates the synthesis of cyclic adenosine monophosphate (cAMP). Cyclic adenosine monophosphate increases the water permeability of the distal convoluted tubule and collecting duct by stimulating the insertion of aquaporin (water) channels in the apical membrane, and thereby increasing water reabsorption and urine concentration, and minimizing urinary water losses. Conversely, as plasma osmolality falls, thirst and ADH release are diminished. This results in decreased renal cAMP generation, decreased water permeability of the distal convoluted tubule and collecting duct, and excretion of a relatively dilute urine that ultimately corrects the body water excess. Marked alterations in water intake combined with perturbations of these various processes often lead to hypernatremia or hyponatremia.
++
Table 12–6 summarizes the xenobiotics that cause hypernatremia. Hypernatremia occasionally results from the parenteral administration of sodium-containing drugs or rapid and excessive oral Na+ intake.1 Oral NaCl and oral sodium citrate were once used as emetics and antiemetics, respectively. As might be expected, they produced severe hypernatremia.20 One case of unintentional fatal hypernatremia resulted from gargling with a supersaturated NaCl solution.99 Similarly, massive ingestion of sodium hypochlorite bleach was associated with hypernatremia.123 Unfortunately intentional salt poisoning is also reported.37
++
++
More commonly, hypernatremia results from relatively electrolyte-free (hypotonic) water losses due to xenobiotics or conditions that cause urinary, GI, and dermal fluid losses.1 Indeed, all fluid losses from the body, except hemorrhage and those from fistulas, are hypotonic (and have the potential to cause hypernatremia). The lack of adequate fluid replacement is a key element in the development of hypernatremia because even the large fluid losses caused by DI or cholera-induced diarrhea will not cause hypernatremia if they are adequately replaced. Thus, in patients with hypernatremia caused by fluid losses, the reason why the losses were not replaced by the patient should always be sought.
++
Xenobiotics that produce significant diarrhea, such as lactulose, cause hypernatremia through unreplaced stool water losses. A similar pathogenesis accounts for the hypernatremia caused by the polyethylene-containing solution used for bowel preparation for colonoscopy.13 Of particular concern is the use of cathartics in the management of poisonings, especially when fluid losses are not anticipated. For example, multiple doses of sorbitol reportedly produce severe hypernatremic dehydration and death in both children and adults.23,44,55,147
++
Significant water loss also occur through the skin. Although diffuse diaphoresis resulting from cocaine or organic phosphorus insecticide toxicity has the potential to produce hypernatremia, this rarely, if ever, occurs. However, application of a remedy containing hyperosmolar povidone–iodine to the skin of burned patients is reported to produce significant water losses and hypernatremia.130
+++
Diagnosis and Treatment
++
The symptoms of significant hypernatremia consist largely of altered mental status ranging from confusion to coma, and neuromuscular weakness that occasionally results in respiratory paralysis. If hypernatremia is associated with Na+ losses and marked ECFV depletion, cardiovascular symptoms, tachycardia, and orthostatic hypotension may be present. Treatment consists of first replacing the Na+ deficit if present (with isotonic fluids such as 0.9% NaCl solution), and then replacing the water deficit. The water deficit is estimated by assuming that the fractional increase in serum [Na+] is equal to the fractional decrease in total body water. Thus, a serum [Na+] that has increased by 10% (from 144 mEq/L to 158 mEq/L) indicates that the water deficit is 10% (3.6 L in a 60-kg person with 36 L of body water).
++
When hypernatremia develops over several hours, for example, as occurs after ingestion or administration of a sodium salt, rapid correction is indicated. However, when hypernatremia develops over several days or when the duration is unknown, slow correction of hypernatremia (over several days) is recommended.1 The adaptation of brain cells to the water deficit (including the gain of intracellular solute including K+, Na+, inositol, glutamate, taurine, and creatine) makes cerebral edema a frequent complication of rapid water replacement. Although some sources suggest that 0.9% NaCl solution is an appropriate replacement fluid regardless of the magnitude of the water deficit, a more refined analysis emphasizes the use of hypotonic fluids to correct hypernatremia in the absence of a significant sodium deficit.1
++
The greatest water losses and therefore the potentially most severe cases of hypernatremia occur during DI, which is always characterized by greater or lesser degrees of hypotonic polyuria. Diabetes insipidus is either characterized as neurogenic, resulting from failure to sense a rising osmolality or from a failure to release ADH, or nephrogenic, resulting from failure of the kidney to respond appropriately to ADH. Although there are many nontoxicologic causes of DI (eg, trauma, tumor, sarcoidosis, vascular, and congenital), xenobiotic-induced DI is also common and is mediated through either central or peripheral mechanisms.
++
Ethanol, opioid antagonists, and α-adrenergic agonists all suppress ADH release.103,104 Lithium,88,133 demeclocycline,132 methoxyflurane,83,92 propoxyphene, foscarnet,107 mesalazine,90 streptozotocin, amphotericin,69 glufosinate,141 lobenzarit,124 rifampin,116 temozolomide,42 and colchicine146 are associated with nephrogenic DI (Table 12–6). In addition, nephrogenic DI is reportedly caused by severe hypokalemia from diuretic use and hypercalcemia from vitamin D poisoning.146 Of all of these xenobiotics, lithium has been the most extensively evaluated. Although polyuria is a common finding with lithium therapy (occurring in 20%–70% of patients on maintenance therapy), the exact incidence of DI and hypernatremia is unknown. Estimates range from 10% to 20% to as high as 80%.
++
Patients with DI complain of polyuria and polydipsia. Urine volumes typically exceed 30 mL/kg/d146 and can be as high as 9 L/d with nephrogenic DI88 and 12 to 14 L/d with neurogenic (central) DI.104 Nocturia, fatigue, and decreased work performance are often noted.146 Neurogenic DI resulting from hypothalamic or pituitary damage is typically associated with other signs of neuroendocrine dysfunction.104
++
After polyuria is confirmed (eg, in adults, by measuring a urine output >200 mL over one hour), the urine osmolality or specific gravity should be measured. The diagnosis of DI is established by the occurrence of dilute urine (urine osmolality <300 mOsm/kg, urine specific gravity <1.010) in the presence of increased serum [Na+] and a serum osmolality greater than 295 mOsm/kg.146 After this determination, a trial of desmopressin, an arginine vasopressin analog, helps to differentiate between neurogenic and nephrogenic DI. If the etiology of the DI is neurogenic, the patient will promptly respond to desmopressin with a decrease in urine output and increase in urine osmolality. In nephrogenic DI, desmopressin will have no significant effect.
++
The initial approach to a hypernatremic patient with DI involves the repletion of the water deficit (as described earlier) and the restoration of electrolyte depletion, if necessary. If a reversible cause for the DI can be established, it should be corrected. Specifically, xenobiotics implicated as the cause of DI should be discontinued or their dose reduced. Patients with neurogenic DI should be maintained on either vasopressin or desmopressin. The latter is preferred because of the lack of vasopressor effects and ease of administration. In the past, patients were occasionally treated with oral medications known to produce SIADH (see later). Patients with nephrogenic DI can be treated with thiazide diuretics,38 prostaglandin inhibitors,33,86 or amiloride, all of which reduce the urine flow rate.16
++
Hyponatremia may be associated with a high, normal, or low serum osmolality. Some patients with myeloma or severe hyperlipidemia exhibit artifactual hyponatremia whenever the measurement technique requires dilution of the serum sample rather than direct measurement by a sodium electrode. These patients have a normal serum osmolality and no symptoms related to their artifactual hyponatremia, and they require no therapy for their “pseudohyponatremia.”
++
Hyperglycemic patients develop hyponatremia because the increase in plasma osmolality caused by hyperglycemia results in a water shift from the intracellular to the extracellular space. The reduction in serum [Na+], which may cause symptoms, is approximately 1.6 mEq/L for every 100 mg/dL increase in serum glucose concentration above normal. The contribution of hyperglycemia to the hyponatremia should be calculated to determine if other causes of hyponatremia should also be sought. All other causes of hyponatremia are associated with a low plasma osmolality. In fact, in the absence of myeloma, hyperlipidemia, and hyperglycemia, the serum osmolality need not be measured in hyponatremic patients and should be assumed to be low.
++
Hyponatremia associated with a low plasma osmolality usually results from water intake in excess of the renal capacity to excrete it. When renal water excretion is normal, very large intake is required to cause hyponatremia. For example, large quantities of water are ingested over a short period of time by people with psychiatric or neurologic disorders such as psychogenic polydypsia.57,121 Xenobiotic-induced water excess comparable to psychogenic polydipsia is quite uncommon. An example occurred during urologic procedures, such as transurethral resection of the prostate (TURP), in which large volumes of irrigation solution are required. Because the wounds were electrically cauterized, these fluids did not contain conductive electrolytes such as sodium. Sorbitol, dextrose, and mannitol were tried as irrigation solutions in an attempt to maintain a normal osmolality, but their optical characteristics were undesirable during the surgery. Thus, irrigation during TURP was performed with glycine-containing solutions. If a large volume of 1.5% glycine (osmolality, 220 mOsm/kg) is absorbed through the prostatic venous plexus, a rapid reduction in serum [Na+] results and will persist until the glycine is metabolized.66,98 Symptoms in these patients are probably a result of several factors: hyponatremia, the glycine itself, and NH3, a glycine metabolite. A similar complication is also described during hysteroscopy.113
++
Rarely, hyponatremia results from the loss of a body fluid with a [Na+] greater than the ECF [Na+] (of 145 mEq/L). This rarely occurs in patients with adrenal insufficiency through hypertonic urinary losses (although increased ADH secretion as a consequence of ECF sodium depletion is probably a more important mechanism; see later discussion). In burn patients, Na+ may be lost directly from the ECF. When treated with topical applications of silver nitrate cream, hyponatremia developed from the diffusion of sodium through permeable skin into the hypotonic dressing.28 Ingestion of licorice that contains glycyrrhizic acid produces a syndrome of hyponatremia, hypokalemia, and hypertension that resembles mineralocorticoid excess. Although the exact mechanism of hyponatremia is debated, one report suggested that a glycyrrhizic acid–induced reduction in 11-β-hydroxysteroid dehydrogenase activity, an enzyme that metabolizes cortisol, could account for the findings.43 Lithium, which is usually associated with DI and hypernatremia, is rarely reported to cause renal sodium wasting and hyponatremia that seems to be unrelated to ADH effects.97
++
Most cases of hyponatremia are caused by water intake in excess of a reduced renal excretory capacity. This reduction in urinary water excretion may be physiologic (as during ECF sodium depletion) or pathologic (in association with kidney, heart, or liver failure).2 Because these conditions are accompanied by alterations in renal sodium handling, signs and symptoms of ECFV depletion, such as postural hypotension, or ECFV excess, such as edema, usually accompany the hyponatremia. Other patients cannot excrete water normally because malignancy or various brain or pulmonary diseases cause ADH secretion.2 In some cases, the tumors are associated with paraneoplastic disease and directly secrete ADH. Xenobiotics, such as diuretics, cause ECFV depletion, but most directly stimulate ADH secretion or augment the renal effects of ADH. Drugs such as the thiazide diuretics cause hyponatremia by several mechanisms, including interference with maximal urinary dilution, and by ADH-induced water retention in response to decreased ECFV.2,46 Patients with excess secretion or action of ADH who have near-normal ECFV have the syndrome of inappropriate antidiuretic hormone (SIADH) secretion. Table 12–7 summarizes these and other causes of hyponatremia.
++
+++
Syndrome of Inappropriate Antidiuretic Hormone (SIADH)
++
The SIADH is characterized by hyponatremia and plasma hypotonicity in the absence of abnormalities of ECFV, or abnormal adrenal, thyroid, or kidney function. Early reviews claimed that SIADH was a disorder of volume overload based largely on evidence of weight gain.103 The consistent absence of edema, however, and the fact that the decrease in serum [Na+] cannot be accounted for by the fluid gain (weight gain) suggest that water retention is only part of the mechanism.82 Urinary Na+ loss and Na+ redistribution from the extracellular to the intracellular space is apparently important as well.
++
There are many nontoxicologic etiologies of SIADH, most of which result from pulmonary or intracranial pathology. These causes include infections, malignancies, and surgery.2,82,104 Table 12–7 summarizes xenobiotics and other causes known to produce SIADH. The antidiabetics, including both the sulfonylurea (eg, chlorpropamide) and biguanide (eg, metformin) classes, produce hyponatremia more commonly than other drugs.102 Their actions are multifactorial and can include both the potentiation of endogenous ADH and the stimulation of ADH release.102 Many psychiatric medications, including the selective serotonin reuptake inhibitors, cyclic antidepressants, and antipsychotics, are implicated in causing SIADH.79,87,137,144 The effects of these drugs are mediated by the complex interactions between the dopaminergic and noradrenergic systems that control ADH release.137 Additional evidence supports a role of serotonin in drug-induced SIADH. Serotonin (specifically 5-HT2 or 5-HT1C agonism) directly stimulates ADH release10 and water intake.71 An important role of serotonin is supported by the occurrence of SIADH with MDMA use.70
++
The clinical presentation of patients with hyponatremia depends on the cause, the absolute serum [Na+], and the rate of decline in serum [Na+]. Patients with associated ECFV excess or depletion present with evidence of altered ECFV, as well as signs and symptoms of the disease that caused the abnormality in ECFV, such as adrenal insufficiency or heart failure.2 Rarely do these patients exhibit symptoms of hyponatremia and hyposmolality of body fluids per se. This is because of the moderate degree of hyponatremia (usually > 130 mEq/L) or the moderate rate of decline in [Na+] or because the loss of Na+ and water limits the development of cerebral edema.85 It is important to note that patients with hyponatremia and a low plasma osmolality (excluding those with primary polydipsia) all have a urinary osmolality that is relatively high regardless of whether they have excess, diminished, or normal ECFV. Consequently, these disorders can only be differentiated by the history, physical examination, and other laboratory test results.
++
Patients with SIADH, if symptomatic, usually present with signs and symptoms of hyponatremia. As noted earlier, the clinical manifestations of hyponatremia are dependent on both the absolute serum [Na+] and its rate of decline.85,104 Whereas chronic slow depression of the [Na+] is usually well tolerated, rapid decreases are associated with symptoms and sometimes catastrophic events. Symptoms include headache, nausea, vomiting, restlessness, disorientation, depression, apathy, irritability, lethargy, weakness, and muscle cramps. In more severe cases, respiratory depression, coma, and seizures develop.
++
The diagnosis of SIADH is based on establishing the presence of hyponatremia, a low serum osmolality, and impaired urinary dilution in the absence of edema, hypotension, hypovolemia, and kidney, adrenal, or thyroid dysfunction.82 As discussed earlier, the presence of any of these clinical findings suggests that another cause of hyponatremia is present. A serum uric acid concentration is helpful in differentiating SIADH from other causes of hyponatremia. In the presence of hyponatremia and impaired urinary dilution, patients with SIADH have hypouricemia, whereas patients exhibiting ECFV excess or depletion characteristically have hyperuricemia.31
++
Treatment of patients with demonstrable ECFV excess or depletion should be directed at the abnormal ECFV and its cause rather than the hyponatremia. In almost all cases, the hyponatremia will improve with correction of the ECFV.2,85 In a similar way, correction of the serum glucose in hyperglycemic patients and the removal of glycine by hemodialysis in patients with the TURP syndrome will correct the serum [Na+]. The rate of correction of the serum [Na+] in these patients is generally not of concern.
++
In patients with SIADH, treatment begins with fluid restriction. Because the goal of this therapy is to establish a negative fluid balance, careful attention to intake and output is required. If an offending xenobiotic is identified, it should be discontinued. Although most cases resolve in 1 to 2 weeks,82,104 SIADH caused by chronic cerebral or pulmonary conditions or by malignancy often persists. If this occurs, therapy with demeclocycline, lithium, or tolvaptan is suggested because severe fluid restriction is often intolerable. Tolvaptan, an oral ADH antagonist with specificity for the V2 receptor, is currently the recommended treatment for chronically hyponatremic patients in whom fluid restriction is unsuccessful because it is easier to titrate with fewer potential side effects than demeclocycline and lithium. In all asymptomatic or mildly symptomatic patients (usually patients with chronic hyponatremia of more than 2 days’ duration), correction should proceed slowly and certainly at a rate less than 8.0 mEq/L during the first 24 hours and less than 16 mEq/L during the first 48 hours. This approach to correction is indicated because if hyponatremia resolution is faster than the reuptake of the solutes (K+, Na+, inositol, glutamate, taurine, and creatine) extruded from brain cells during the development of hyponatremia, brain shrinkage will occur with disruption of the blood–brain barrier. This effect is responsible for the osmotic demyelination syndrome (ODS). The ODS, which is associated with central pontine and extrapontine myelinolysis, often has a delayed onset of 2 to 6 days and causes irreversible brain damage and death in 50% and 15% of patients, respectively.2,12,139 Risk factors for ODS, in addition to too rapid correction of chronic hyponatremia, include serum [Na+] below 115 mEq/L and patients who are elderly, hypokalemic, alcoholic, or malnourished. In asymptomatic patients, as described earlier, water restriction is recommended and usually sufficient, but occasionally tolvaptan (an ADH V2 receptor antagonist) is appropriate. When hyponatremia is associated with life-threatening clinical presentations, including respiratory depression, altered mental status, seizures, or coma, careful infusion of hypertonic 3% saline ([Na+] = 513 mEq/L) (eg, 1–2 mL/kg/h), with or without furosemide, is indicated.2,82 Alternatively, conivaptan (another ADH V2 receptor antagonist) can be administered intravenously as an initial bolus of 20 mg followed by a continuous infusion of 40 mg/d for no more than 4 days.62,65 In these symptomatic patients, the goal is to increase the serum [Na+] 1 mEq/L/h (5–6 mEq/L over 6 hours) or until life-threatening symptoms resolve.2,85 After this initial correction and improvement of symptoms, the serum [Na+] should be increased at a rate less than 0.3 mEq/L/h, preferably by water restriction alone. Formulas are available to help calculate the rate of correction of hyponatremia.2 Equation 12–8 is the equation we prefer.
++
When 1 L of fluid is infused:
++
++
where the infusate [Na+] in mEq/L equals:
++
+++
XENOBIOTIC-INDUCED ELECTROLYTE ABNORMALITIES
++
Xenobiotic-induced alterations in serum [K+] are potentially more serious than alterations in other electrolyte concentrations because of the critical role of potassium in a variety of homeostatic processes, most importantly, muscle strength and cardiac function. The total body potassium content of an average adult is about 54 mEq/kg, of which only 2% is located in the intravascular space. The large intracellular store of potassium is maintained by a variety of systems, the most important of which is membrane Na+,K+-adenosine triphosphatase (ATPase). The relationship between total body stores and serum [K+] is not linear, and small changes in the total body potassium often result in dramatic alterations in serum concentrations and, more importantly, in the ratio of extracellular to intracellular [K+].
++
People eating a Western diet ingest 50 to 150 mEq/d of potassium, approximately 90% of which is subsequently eliminated in the urine. The body has 2 major defenses against a potassium load: acutely, potassium is transferred into cells; chronically, potassium is excreted in the urine by decreased proximal tubular reabsorption and increased distal tubular secretion (to a maximum of 600–700 mEq/d).22 After a meal, K+ transfers into the intracellular space through insulin and catecholamine-mediated uptake of potassium in liver and muscle cells.122 Renal potassium excretion is primarily modulated by the renin–angiotensin–aldosterone system. In addition, the GI absorption of potassium decreases as the serum [K+] increases.
++
Hypokalemia results from decreased oral intake, GI losses caused by repeated vomiting or diarrhea, urinary losses through increased K+ secretion or decreased reabsorption, and processes that shift potassium into the intracellular compartment.20,22,148 Table 12–8 summarizes the xenobiotics and other causes commonly associated with hypokalemia.
++
++
The neuromuscular manifestations of hypokalemia are reviewed elsewhere.80 Patients with hypokalemia are often asymptomatic when the decrease in serum [K+] is mild (serum concentrations of 3.0–3.5 mEq/L). Occasionally, hypokalemia interferes with renal concentrating mechanisms, and polyuria is noted. More significant potassium deficits (serum concentrations of 2.0–3.0 mEq/L) cause generalized malaise and weakness. As the [K+] falls to less than 2 mEq/L, weakness becomes prominent, and areflexic paralysis and respiratory failure occur, often necessitating intubation and mechanical ventilation.80 Rhabdomyolysis also occurs. These neuromuscular manifestations are so prominent that they are potentially erroneously attributed to a neuromuscular syndrome such as Guillain-Barré. Other clinical findings associated with hypokalemia include GI hypoperistalsis (ileus); manifestations of cardioactive steroid toxicity; worsening hyperglycemia in patients with diabetes; and the symptoms and signs of the metabolic abnormalities that often accompany hypokalemia, such as hyponatremia, metabolic acidosis, or alkalosis.148
++
Electrocardiographic changes also are common, even with mild potassium depletion, although the absence of ECG changes should never be used to exclude significant hypokalemia. Common ECG findings of hypokalemia include depression of the ST segment, decreased T-wave amplitude, and increased U-wave amplitude (Chap. 15). These findings may herald life-threatening rhythm disturbances, particularly polymorphic ventricular tachycardia.148
++
Treatment of hypokalemia involves discontinuing or removing the offending xenobiotic and correcting the potassium deficit. Potassium supplementation should be given, either orally or intravenously as acceptable. The choice of potassium salt should be based on the associated acid–base abnormality, if present. Thus, potassium chloride is preferred when metabolic alkalosis is present, and another salt of potassium (eg, potassium citrate or potassium bicarbonate) is preferable when metabolic acidosis is present.148 Potassium phosphate should be used as part of the K+ supplement when hypophosphatemia is also present, as occurs in diabetic ketoacidosis, or when hyperchloremia is present. Hypomagnesemia, which either causes or accompanies hypokalemia (eg, in diuretic-induced hypokalemia), must be corrected because this abnormality often prevents successful potassium replacement.
++
The debate over the maximum safe infusion rate for IV potassium is summarized elsewhere.84,148 Based on experience with more than 1300 infusions, one group concluded that under intensive care monitoring, IV administration of 20 mEq/h (by central or peripheral vein) is well tolerated. They also found that each 20 mEq of potassium administered resulted in an average increase in serum [K+] of 0.25 mEq/L. Under most circumstances we agree with these recommendations, but acknowledge that others have used significantly larger doses (up to 100 mEq/h) in unique life-threatening circumstances.32
++
Hyperkalemia results from decreased urinary elimination (renal insufficiency, potassium-sparing diuretics, hypoaldosteronism), increased intake (either orally or intravenously), or redistribution from tissue stores.20,22 The last mechanism is of major toxicologic importance. Overdoses of both cardioactive steroids (Chap. 62) and β-adrenergic antagonists (Chap. 59) cause hyperkalemia by promoting net potassium release from intracellular reservoirs. Presumably because of other protective mechanisms, overdose with a β-adrenergic antagonist produces only a moderate rise in serum [K+] (usually to 5.0–5.5 mEq/L). By contrast, a similar rise in serum [K+] as a consequence of blockade of the Na+,K+-ATPase pump during acute cardioactive steroid toxicity may be lethal (Chap. 62). This suggests that hyperkalemia per se is not the cause of the lethality of cardioactive steroid toxicity. Thus, the focus of therapy should involve efforts to neutralize or eliminate the cardioactive steroid rather than reduce the serum [K+].18 Table 12–8 lists xenobiotics and other causes of hyperkalemia.
++
After oral overdoses of potassium salts, patients usually complain of nausea and vomiting. Ileus, intestinal irritation, bleeding, and perforation reportedly complicate the clinical course.126,127 In the absence of potassium ingestion, GI symptoms of hyperkalemia are usually very mild. Neuromuscular manifestations include weakness with an ascending flaccid paralysis and respiratory compromise, with intact sensation and cognition.80,94,114 The similarity of these signs and symptoms to those associated with hypokalemia is striking, suggesting that hyperkalemia should only be diagnosed with certainty by laboratory measurement.
++
The cardiac manifestations of hyperkalemia are distinct, prominent, and life threatening. ECG patterns progress through characteristic changes.127 Although the progression of ECG changes is reproducible, there is great individual variation with respect to the serum [K+] at which these ECG findings occur. Initially, the only ECG finding is the presence of tall, peaked T waves. As the serum [K+] concentration increases, the QRS complex tends to blend into the T waves, the P-wave amplitude decreases, and the PR interval becomes prolonged. Next, the P wave is lost, and ST-segment depression occurs. Finally, the distinction between the S and T waves becomes blurred, and the ECG takes on a sine wave configuration (Chap. 16). Hemodynamic instability and cardiac arrest can result. As the patient’s serum [K+] falls with therapy, these ECG changes resolve in a reverse fashion.
++
The treatment of severe hyperkalemia focuses on methods to (1) reverse the ECG effects, (2) transfer K+ to the intracellular space, and (3) enhance K+ elimination. Pharmacologic interventions, extensively discussed elsewhere,126 are summarized here. Calcium salts (eg, 10–20 mEq administered intravenously) (Antidotes in Depth: A29) work almost immediately to protect the myocardium against the effects of hyperkalemia, although it does not reduce the serum [K+]. However, a potentially life-threatening interaction occurs when the patient with cardioactive steroid toxicity is given calcium salts (Chap. 62).
++
The administration of insulin (and dextrose to prevent hypoglycemia unless hyperglycemia is present), sodium bicarbonate, or inhalation of a β-adrenergic agonist all stimulate potassium entry into cells.8 They reduce the serum [K+] over approximately 30 minutes, but potassium begins to reenter the extracellular space over the next several hours. Caution is advised, when using insulin in patients with chronic kidney disease because hypoglycemia can be prolonged as a result reduced insulinase function. Cationic exchange resins, such as Na+ polystyrene sulfonate and patiromer, nonabsorbable polymers that binds potassium in exchange for calcium take longer to reduce the serum [K+] as they enhance GI potassium loss.14,63,81 Hemodialysis or peritoneal dialysis are the most efficient means of rapidly reducing total body potassium stores, especially when significant renal impairment is present.
++
Calcium is the most abundant mineral in the body, and 98% to 99% is located in bone. Approximately half of the remaining 1% to 2% of calcium in the body is bound to plasma proteins (mostly albumin), and most of the rest is complexed to various anions, with free, ionized calcium representing a very small fraction of extraosseous stores. The serum [Ca2+] is maintained through interactions between dietary intake and renal elimination, modulated by vitamin D activity, parathyroid hormone, and calcitonin. More extensive discussions of calcium physiology are found elsewhere.6,105
++
Xenobiotic-induced hypercalcemia is uncommon and usually caused by increased dietary calcium as a result of milk or antacid usage, calcium supplements, or a decrease in its renal excretion such as occurs with thiazide use.6,20 Cholecalciferol, available as a rodenticide, can increase the serum [Ca2+] by increasing its release from bone, increasing GI absorption, and decreasing renal elimination. Vitamin D toxicity from excessive vitamin or milk intake also can cause hypercalcemia.76 Table 12–9 lists other causes of hypercalcemia.
++
++
Symptoms of hypercalcemia consist of lethargy, muscle weakness, nausea, vomiting, and constipation. Life-threatening manifestations include complications from altered mental status such as aspiration pneumonia and cardiac dysrhythmias (Chap. 15). Treatment of clinically significant hypercalcemia focuses on removing the offending xenobiotic when possible, decreasing GI absorption by administering a binding agent, increasing distribution into bone with a bisphosphonate (onset 1–2 days), and enhancing renal excretion through forced diuresis with IV 0.9% NaCl solution and furosemide (onset 4–6 hours).6,140 Hemodialysis is often required when significant renal impairment is present.
++
Xenobiotics more commonly cause hypocalcemia than hypercalcemia. Minor, usually clinically insignificant, decreases in serum [Ca2+] can occur in association with anticonvulsant and aminoglycoside therapy.20 Severe, life-threatening hypocalcemia can occur, however, from ethylene glycol poisoning (Chap. 106) or as a manifestation of fluoride toxicity from either fluoride salts or hydrofluoric acid (Chap. 103).39,142 Calcium complex formation with fluoride or oxalate ions is responsible for the rapid development of hypocalcemia in these settings. Similar effects occur with excess phosphate145 or citrate93,143 intake. This mechanism (calcium complex formation) decreases the ionized [Ca2+] but may or may not reduce the measured total serum [Ca2+]. Other xenobiotics that produce hypocalcemia decrease absorption, enhance renal loss, or stimulate calcium entry into cells (Table 12–9).
++
Symptoms of hypocalcemia consist largely of neuromuscular findings, including paresthesias, cramps, carpopedal spasm, tetany, and seizures. Although ECG abnormalities are common (Chap. 15), life-threatening dysrhythmias are rare. Treatment strategies focus on calcium replacement. When hypomagnesemia or hyperphosphatemia is present, these abnormalities should be corrected, or calcium replacement will likely fail.
++
Magnesium is the fourth most abundant cation in the body (after Ca2+, Na+, and K+), with a normal total body store of about 2000 mEq in a 70-kg person.117 Approximately 50% of magnesium is stored in bone, with most of the remainder distributed in the soft tissues. Because only approximately 1% to 2% of magnesium is located in the ECF, serum concentrations correlate poorly with total body stores.118 Magnesium homeostasis is maintained through dietary intake and renal and GI excretion modulated by hormonal effects.5
++
Clinically significant hypermagnesemia is uncommon in the absence of kidney failure except when massive doses of magnesium salts overwhelm renal excretory mechanisms. This has been reported with inadvertent IV infusion,19,24,68,101 urologic procedures involving irrigation with magnesium salts,45,77 and ingestion of large quantities of magnesium-containing antacids95 or cathartics.52,56 Iatrogenic overdose was formerly of concern when magnesium-containing cathartics were routinely used in poison management.54,78,135 In a series of poisoned patients, a single oral dose of a magnesium-containing cathartic (30 g of magnesium sulfate) failed to produce any demonstrable rise in serum [Mg2+].136 However, patients who received 3 oral doses of magnesium sulfate over 8 hours had a significant increase in their serum [Mg2+].136 Thus, the potential for iatrogenic toxicity exists, mandating cautious use of magnesium-containing cathartics, especially in patients with renal insufficiency. Table 12–10 lists xenobiotic causes of hypermagnesemia.
++
++
The symptoms of hypermagnesemia correlate with serum concentrations but depend somewhat on the rate of increase and host factors. At serum [Mg2+] of about 3 to 10 mEq/L, patients feel weak, nauseated, flushed, and thirsty. Bradycardia, a widened QRS complex on ECG, hypotension, and decreased deep tendon reflexes are typically noted. As the [Mg2+] increases, hypoventilation, muscle paralysis, and ventricular dysrhythmias occur. Serum [Mg2+] greater than 10 mEq/L, especially those concentrations greater than 15 mEq/L, often cause death.
++
Hypermagnesemia is a life-threatening disorder. When significant neuromuscular or ECG manifestations are noted, administration of 5 to 20 mEq of Ca2+ should be administered by slow intravenous infusion to competitively antagonize the effects of the magnesium ion.6,61 (Antidotes in Depth: A32) Further therapy should focus on enhancing renal excretion by administering fluids and loop diuretics such as furosemide.61 In the presence of renal failure or inadequate renal excretion, hemodialysis will rapidly correct hypermagnesemia.
++
Xenobiotic-induced hypomagnesemia is common but rarely life threatening. Renal losses (caused by diuretics), GI losses (caused by ethanol), intracellular shifts caused by insulin91 or β-adrenergic agonists, and complex formation (by fluoride or phosphate) are common.5,6 Table 12–10 lists these and other xenobiotic causes of hypomagnesemia. Of note, many causes of hypomagnesemia also cause hypokalemia and hypocalcemia.5 Therefore, when hypomagnesemia is suspected or discovered, the presence of other electrolyte abnormalities should be sought.
++
The symptoms of hypomagnesemia are lethargy, weakness, fatigue, neuromuscular excitation (tremor and hyperreflexia), nausea, and vomiting.29,35 Dysrhythmias are reported, especially in patients treated with cardioactive steroids. Signs and symptoms consistent with hypocalcemia and hypokalemia also may be present.
++
Treatment involves removing the offending xenobiotic (if it can be identified) and restoring magnesium balance. Although either oral or parenteral supplementation is usually acceptable for mild hypomagnesemia, parenteral therapy is required when significant clinical manifestations are present. When oral therapy is indicated, a normal diet or magnesium oxide, magnesium chloride, or magnesium lactate in divided doses (magnesium 20–100 mEq/d) will often correct the hypomagnesemia.5,6 When hypomagnesemia is severe or symptomatic, and kidney function is normal, 16 mEq (2 g) of magnesium sulfate can be given intravenously over several minutes to a maximum of 1 mEq/kg of magnesium in a 24-hour period.6,29,75 During any continuous magnesium infusion, frequent serum [Mg2+] determinations should be obtained and the presence of reflexes documented. If hyporeflexia occurs, the magnesium infusion should be discontinued and the patient assessed and treated for respiratory muscle weakness if present.
++
The management of poisoned patients must include an evaluation of their fluid, electrolyte, and acid–base status.
Developing a stepwise approach to the evaluation of patients with a high anion gap metabolic acidosis is essential.
The osmol gap can help confirm a suspicion of toxic alcohol poisoning but cannot exclude the diagnosis.
Disorders of sodium (water) are common manifestations of xenobiotic exposure. Because fluid and electrolyte abnormalities result from both poisoning and the therapy of poisoned patients, reassessment and monitoring are essential to ensure a good clinical outcome.
1. +
Adrogue
HJ, Madias
NE. Hypernatremia.
N Engl J Med. 2000;342:1493–1499.
[PubMed: 10816188]
2. +
Adrogue
HJ, Madias
NE. Hyponatremia.
N Engl J Med. 2000;342:1581–1589.
[PubMed: 10824078]
3. +
Adrogue
HJ, Madias
NE. Management of life-threatening acid-base disorders. First of two parts.
N Engl J Med. 1998;338:26–34.
[PubMed: 9414329]
4. +
Adrogue
HJ, Madias
NE. Management of life-threatening acid-base disorders. Second of two parts.
N Engl J Med. 1998;338:107–111.
[PubMed: 9420343]
6. +
Agus
ZS, et al. Disorders of calcium and magnesium homeostasis.
Am J Med. 1982;72:473–488.
[PubMed: 7036737]
7. +
Albert
MS, et al. Quantitative displacement of acid-base equilibrium in metabolic acidosis.
Ann Intern Med. 1967;66:312–322.
[PubMed: 6016545]
8. +
Allon
M, et al. Nebulized
albuterol for acute hyperkalemia in patients on hemodialysis.
Ann Intern Med. 1989;110:426–429.
[PubMed: 2919849]
9. +
Ammar
KA, Heckerling
PS. Ethylene glycol poisoning with a normal anion gap caused by concurrent ethanol ingestion: importance of the osmolal gap.
Am J Kidney Dis. 1996;27:130–133.
[PubMed: 8546127]
10. +
Anderson
IK, et al. Central administration of 5-HT activates 5-HT1A receptors to cause sympathoexcitation and 5-HT2/5-HT1C receptors to release
vasopressin in anaesthetized rats.
Br J Pharmacol. 1992;107:1020–1028.
[PubMed: 1467825]
11. +
Andrews
TJ, et al. Euglycemic diabetic ketoacidosis with elevated acetone in a patient taking a sodium-glucose cotransporter-2 (SGLT2) inhibitor.
J Emerg Med. 2017;52:223–226.
[PubMed: 27717592]
12. +
Ayus
JC, et al. Treatment of symptomatic hyponatremia and its relation to brain damage. A prospective study.
N Engl J Med. 1987;317:1190–1195.
[PubMed: 3309659]
13. +
Ayus
JC, et al. Fatal dysnatraemia caused by elective colonoscopy.
BMJ. 2003;326:382–384.
[PubMed: 12586675]
15. +
Batlle
DC, et al. The use of the urinary anion gap in the diagnosis of hyperchloremic metabolic acidosis.
N Engl J Med. 1988;318:594–599.
[PubMed: 3344005]
17. +
Berl
T, et al. Clinical disorders of water metabolism.
Kidney Int. 1976;10:117–132.
[PubMed: 7703]
18. +
Bismuth
C, et al. Hyperkalemia in acute digitalis poisoning: prognostic significance and therapeutic implications.
Clin Toxicol. 1973;6:153–162.
[PubMed: 4715199]
19. +
Bourgeois
FJ, et al. Profound hypotension complicating magnesium therapy.
Am J Obstet Gynecol. 1986;154:919–920.
[PubMed: 3963084]
20. +
Brass
EP, Thompson
WL. Drug-induced electrolyte abnormalities.
Drugs. 1982;24:207–228.
[PubMed: 6751789]
21. +
Brenner
BE. Clinical significance of the elevated anion gap.
Am J Med. 1985;79:289–296.
[PubMed: 4036980]
22. +
Brown
RS. Extrarenal potassium homeostasis.
Kidney Int. 1986;30:116–127.
[PubMed: 3528616]
23. +
Caldwell
JW, et al. Hypernatremia associated with cathartics in overdose management.
West J Med. 1987;147:593–596.
[PubMed: 3424825]
24. +
Cao
Z, et al. Acute hypermagnesemia and respiratory arrest following infusion of MgSO4 for tocolysis.
Clin Chim Acta. 1999;285:191–193.
[PubMed: 10481936]
25. +
Carlisle
EJ, et al. Glue-sniffing and distal renal tubular acidosis: sticking to the facts. Journal of the American Society of Nephrology: J Am Soc Nephrol. 1991;1:1019–1027.
26. +
Carstairs
SD, et al. Contribution of serum ethanol concentration to the osmol gap: a prospective volunteer study.
Clin Toxicol (Phila). 2013;51:398–401.
[PubMed: 23641935]
27. +
Clement
M, Senior
P. Euglycemic diabetic ketoacidosis with
canagliflozin: not-so-sweet but avoidable complication of sodium-glucose cotransporter-2 inhibitor use.
Can Fam Physician. 2016;62:725–728.
[PubMed: 27629668]
29. +
Cronin
RE, Knochel
JP. Magnesium deficiency.
Adv Intern Med. 1983;28:509–533.
[PubMed: 6340431]
30. +
Darchy
B, et al. Delayed admission for ethylene glycol poisoning: lack of elevated serum osmol gap.
Intensive Care Med. 1999;25:859–861.
[PubMed: 10447547]
31. +
Decaux
G, et al. Uric acid, anion gap and urea concentration in the diagnostic approach to hyponatremia.
Clin Nephrol. 1994;42:102–108.
[PubMed: 7955571]
32. +
DeFronzo
RA, Bia
M. Intravenous potassium chloride therapy. 1981;245:2446.
34. +
DiNubile
MJ. The increment in the anion gap: overextension of a concept?
Lancet. 1988;2:951–953.
[PubMed: 2902392]